Your Own Private AI on a $798 Laptop: 2B Model, 91% Pass
We built an AI router that keeps 89% of queries free and on-device. 35 tests. Real numbers.
Educational AI demos, tools, and experimental builds by NosisTech. Published for learning and documentation only, not as production systems, implementation guidance, or professional advice.
We built an AI router that keeps 89% of queries free and on-device. 35 tests. Real numbers.
A $798 gaming laptop with a 4GB GPU runs private local AI and automatically escalates the hard stuff to the cloud. Here’s the test data.
We installed a local AI on a $798 gaming laptop and ran 71 real-world tasks.
You changed the context to 64k and Hermes still rejects it. Here is the parallel slots problem nobody is writing about yet.
How I turned a 12GB laptop GPU into a 114 tok/s local AI agent with MTP, Hermes, and three config changes.
I started this project a while ago with a simple idea: build a personal AI assistant that I could talk to, that would remember our conversations, and that would never send my data to a third party. I named it Korvin. It now runs on a $5 VPS, answers me through Telegram voice messages, and … Read more
Installed Langflow Desktop, built an RFP analyzer by connecting four visual nodes, and called it via API.
Building multi-role marketing agent that enforces brand rules automatically, rejecting non-compliant drafts and retrying until the output passes.
Rebuild the three core pillars of the Phidata agent framework from scratch in pure Python. Memory, knowledge, and tools working together.
AI agent that reads invoice images, extracts structured data, and automatically flags low-confidence results for human review.